서브메뉴
검색
Data-Driven Connected Cruise Control for Mixed Traffic Environments
Data-Driven Connected Cruise Control for Mixed Traffic Environments
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105234
- ISBN
- 9798291567586
- DDC
- 621
- 저자명
- Shen, Minghao.
- 서명/저자
- Data-Driven Connected Cruise Control for Mixed Traffic Environments
- 발행사항
- [Sl] : University of Michigan, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 124 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-03, Section: B.
- 주기사항
- Advisor: Orosz, Gabor.
- 학위논문주기
- Thesis (Ph.D.)--University of Michigan, 2025.
- 초록/해제
- 요약Modern transportation systems are undergoing a significant transformation toward higher level of automation and broader penetration of connectivity. This shift is driven by the promise of improved safety, enhanced traffic efficiency, and reduced energy consumption. However, realizing these benefits in practice faces a fundamental challenge: the presence of mixed traffic environments where connected and automated vehicles (CAVs) must coexist with human-driven vehicles (HVs), non-connected automated vehicles (AV), and connected human-driven vehicles. In such transitional settings, the assumptions of full connectivity and high-level cooperation no longer hold, necessitating robust and scalable control policies that can perform effectively under partial observability and sparse connectivity. This dissertation presents a comprehensive framework for data-driven Connected Cruise Control (CCC), aiming to leverage vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) communications to improve longitudinal vehicle control under realistic constraints. The work systematically develops both reactive and predictive CCC algorithms that incorporate beyond-line-of-sight information from connected vehicles, even in the presence of unknown or stochastic behavior from other traffic participants. The first part of the dissertation focuses on reactive CCC, where a feedback control strategy is formulated to synchronize vehicle speeds and enhance string stability by responding to multiple leading vehicles. A novel spectral analysis framework is introduced, which enables data-driven controller optimization by modeling traffic fluctuations as stationary stochastic processes. This approach provides a surrogate model for energy consumption, allowing closed-form characterization of optimal controller parameters based on the spectral properties of traffic data. The theoretical results are validated through both synthetic simulations and experimental vehicle trajectory datasets. The second part introduces a predictive CCC framework based on model predictive control (MPC). This controller utilizes connectivity information to predict future motions of nearby vehicles, including those beyond direct sensor range, and optimizes control inputs accordingly. Extensions to estimate the number and influence of hidden vehicles are developed to enhance robustness in partially observable settings. Simulation studies show that predictive CCC can significantly outperform conventional adaptive cruise control in terms of energy efficiency and ride comfort. Building upon these foundational results, the final part of the dissertation advances a data-driven predictive control architecture that bypasses the need for precise vehicle models. Instead, it uses system identification tools grounded in behavioral theory of linear time-invariant (LTI) systems. A key contribution is the introduction of a memory sketching technique, which enables real-time implementation by compressing incoming data streams into fixed-size summaries. This drastically reduces computational complexity and memory usage without sacrificing prediction accuracy, thus making the approach scalable for real-world deployment. Together, the contributions of this dissertation demonstrate a cohesive and scalable approach to integrating data-driven techniques with connected vehicle control. By addressing the challenges posed by sparse connectivity, uncertain driver behavior, and real-time implementation, this work lays a foundation for practical connected cruise control systems that are ready for deployment in today's mixed traffic environments. The proposed methods not only improve energy efficiency and traffic flow but also represent a significant step toward the broader vision of intelligent, cooperative, and sustainable transportation systems.
- 일반주제명
- Mechanical engineering
- 일반주제명
- Engineering
- 일반주제명
- Automotive engineering
- 기타저자
- University of Michigan Mechanical Engineering
- 기본자료저록
- Dissertations Abstracts International. 87-03B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017359906
■00520260202105234
■006m o d
■007cr#unu||||||||
■020 ▼a9798291567586
■035 ▼a(MiAaPQ)AAI32271925
■035 ▼a(MiAaPQ)umichrackham006401
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a621
■1001 ▼aShen, Minghao.
■24510▼aData-Driven Connected Cruise Control for Mixed Traffic Environments
■260 ▼a[Sl]▼bUniversity of Michigan▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a124 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-03, Section: B.
■500 ▼aAdvisor: Orosz, Gabor.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2025.
■520 ▼aModern transportation systems are undergoing a significant transformation toward higher level of automation and broader penetration of connectivity. This shift is driven by the promise of improved safety, enhanced traffic efficiency, and reduced energy consumption. However, realizing these benefits in practice faces a fundamental challenge: the presence of mixed traffic environments where connected and automated vehicles (CAVs) must coexist with human-driven vehicles (HVs), non-connected automated vehicles (AV), and connected human-driven vehicles. In such transitional settings, the assumptions of full connectivity and high-level cooperation no longer hold, necessitating robust and scalable control policies that can perform effectively under partial observability and sparse connectivity. This dissertation presents a comprehensive framework for data-driven Connected Cruise Control (CCC), aiming to leverage vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) communications to improve longitudinal vehicle control under realistic constraints. The work systematically develops both reactive and predictive CCC algorithms that incorporate beyond-line-of-sight information from connected vehicles, even in the presence of unknown or stochastic behavior from other traffic participants. The first part of the dissertation focuses on reactive CCC, where a feedback control strategy is formulated to synchronize vehicle speeds and enhance string stability by responding to multiple leading vehicles. A novel spectral analysis framework is introduced, which enables data-driven controller optimization by modeling traffic fluctuations as stationary stochastic processes. This approach provides a surrogate model for energy consumption, allowing closed-form characterization of optimal controller parameters based on the spectral properties of traffic data. The theoretical results are validated through both synthetic simulations and experimental vehicle trajectory datasets. The second part introduces a predictive CCC framework based on model predictive control (MPC). This controller utilizes connectivity information to predict future motions of nearby vehicles, including those beyond direct sensor range, and optimizes control inputs accordingly. Extensions to estimate the number and influence of hidden vehicles are developed to enhance robustness in partially observable settings. Simulation studies show that predictive CCC can significantly outperform conventional adaptive cruise control in terms of energy efficiency and ride comfort. Building upon these foundational results, the final part of the dissertation advances a data-driven predictive control architecture that bypasses the need for precise vehicle models. Instead, it uses system identification tools grounded in behavioral theory of linear time-invariant (LTI) systems. A key contribution is the introduction of a memory sketching technique, which enables real-time implementation by compressing incoming data streams into fixed-size summaries. This drastically reduces computational complexity and memory usage without sacrificing prediction accuracy, thus making the approach scalable for real-world deployment. Together, the contributions of this dissertation demonstrate a cohesive and scalable approach to integrating data-driven techniques with connected vehicle control. By addressing the challenges posed by sparse connectivity, uncertain driver behavior, and real-time implementation, this work lays a foundation for practical connected cruise control systems that are ready for deployment in today's mixed traffic environments. The proposed methods not only improve energy efficiency and traffic flow but also represent a significant step toward the broader vision of intelligent, cooperative, and sustainable transportation systems.
■590 ▼aSchool code: 0127.
■650 4▼aMechanical engineering
■650 4▼aEngineering
■650 4▼aAutomotive engineering
■653 ▼aData-driven control
■653 ▼aConnected and automated vehicle
■653 ▼aHuman-driven vehicles
■653 ▼aAutomated vehicles
■653 ▼aConnected Cruise Control
■690 ▼a0548
■690 ▼a0537
■690 ▼a0540
■71020▼aUniversity of Michigan▼bMechanical Engineering.
■7730 ▼tDissertations Abstracts International▼g87-03B.
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17359906▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


